alma science pipeline quickstart guide
TRANSCRIPT
1
ALMA, an international astronomy facility, is a partnership of Europe, North America and East Asia in cooperation with the Republic of Chile.
ALMA Science Pipeline Quickstart Guide
Doc 2.13, ver. 1.0 | September 2014
2
For further information or to comment on this document, please contact your regional Helpdesk through the ALMA User Portal at www.almascience.org. Helpdesk tickets will be directed to the appropriate ALMA Regional Center at ESO, NAOJ or NRAO.
Version Date Editors
1.0 September 2014 Pipeline Team
In publications, please refer to this document as:
ALMA Pipeline Team, 2014, ALMA Science Pipeline Quickstart Guide, Version 1.0, ALMA
3
Table of contents
1 Overview ................................................................................................................................................. 4
2 Obtaining the Pipeline ........................................................................................................................ 4
3 Pipeline-‐related Documentation .................................................................................................... 4
4 Producing Calibrated Measurement Sets from ALMA Pipeline Deliveries: scriptForPI.py ............................................................................................................................................... 4
5 The Pipeline Interferometry Calibration Script: casa_pipescript.py .................................. 5 5.1 Running casa_pipescript.py from ALMA Delivery Packages ......................................................... 5 5.2 Modifying the Pipeline Run ...................................................................................................................... 6 5.2.1 Note on the Pipeline Mode .................................................................................................................................... 6 5.2.2 Removing ASDMs from the processing ........................................................................................................... 7 5.2.3 Introducing additional flagging .......................................................................................................................... 7 5.2.4 Editing calibrator fluxes used by the Pipeline .............................................................................................. 7
6 The Pipeline WebLog .......................................................................................................................... 7 6.1 Navigation ...................................................................................................................................................... 8 6.2 Home Page ..................................................................................................................................................... 8 6.2.1 Observation Summary ............................................................................................................................................ 8
6.3 By Topic .......................................................................................................................................................... 9 6.4 By Task ............................................................................................................................................................ 9 6.4.1 Task sub-‐pages ........................................................................................................................................................... 9
6.5 Selecting and filtering plot displays .................................................................................................... 10 6.6 Weblog Quality Assessment (QA) Scoring ......................................................................................... 11
4
1 Overview The ALMA Science Pipeline is being developed for the automated processing of ALMA interferometry and single-‐dish data, and for the combination of data taken using multiple ALMA arrays. At present, the Pipeline is commissioned for the calibration of the majority of ALMA interferometry datasets and for the provision of diagnostic calibrator images. However, data combination from multiple arrays, science target imaging and single-‐dish processing are not yet commissioned. This document describes how to obtain the ALMA Pipeline, how to use it to calibrate ALMA data and use of the Pipeline WebLog, the set of plots output by the Pipeline.
The Pipeline calibrates and images ALMA data automatically using Pipeline tasks. Pipeline tasks use CASA tasks wherever possible to perform the data reduction, and Pipeline tasks can be viewed and executed within CASA in exactly the same way as CASA tasks. For example, in a CASA version also containing the Pipeline, to view the inputs possible for Pipeline task hifa_importdata type: inp hifa_importdata.
The Pipeline is data-‐driven i.e. the characteristics of each dataset drives the calibration and imaging strategy (the Pipeline Heuristics). During the Pipeline run, information on which calibration tables should be used etc are stored in the Pipeline Context.
2 Obtaining the Pipeline The Pipeline can be obtained by downloading a version of CASA that also contains the Pipeline. This is available, along with installation instructions, from the Documents & Tools Section of the ALMA Science Portal: ( http://www.almascience.org/). If any issues are encountered with Pipeline installation, please contact the ALMA Helpdesk via the link on the ALMA Science Portal.
3 Pipeline-‐related Documentation The User documentation currently relating to Pipeline, available from the ALMA Science Portal, is:
• ALMA Science Pipeline Quickstart Guide: Overview of the ALMA Science Pipeline (this document)
• ALMA Science Pipeline Reference Manual: Description of individual Pipeline tasks
• ALMA QA2 Data Products for Cycle 2: Description of files included in ALMA deliveries
4 Producing Calibrated Measurement Sets from ALMA Pipeline Deliveries: scriptForPI.py
If the Pipeline was used during the ALMA Quality Assurance Process, this will be noted in the README file of an ALMA delivery. In order to convert ALMA raw data into calibrated Measurement Sets, for example before re-‐imaging the data, then the script scriptForPI.py should be used. scriptForPI.py directly applies calibration and flagging tables determined during the ALMA Quality Assurance process to raw data to create calibrated measurement sets. The instructions for using this script will be in the README file of the ALMA delivery.
5
Using scriptForPI.py is the recommended and fastest method of obtaining calibrated ALMA data from the delivery. For deliveries that used the ALMA Pipeline, scriptForPI.py must be run from within a version of CASA that contains the Pipeline.
5 The Pipeline Interferometry Calibration Script: casa_pipescript.py To make any alterations to the calibration and flagging performed during the ALMA Quality Assurance Process, data have to be fully re-‐calibrated using the script casa_pipescript.py (instead of using scriptForPI.py). casa_pipescript.py is a python script for performing full data re-‐calibration using the Pipeline (typical script shown in Figure 1). Note that the Pipeline works on completed Observing Unit Sets (defined in the ALMA Technical Handbook), such that multiple executions of the same Scheduling Block (execution blocks, ASDMs) can be calibrated using a single script to produce a calibrated Measurement Set per ASDM. Information on individual Pipeline tasks is in the ALMA Science Pipeline Reference Manual.
5.1 Running casa_pipescript.py from ALMA Delivery Packages
To reproduce the Pipeline calibration performed as part of the ALMA Quality Assurance Process: • Copy casa_pipescript.py from the script directory to the raw directory • Copy flux.csv and *flagtemplate.txt from the calibration directory to the raw directory
In the raw directory: • Make sure the naming of the raw ALMA data is consistent with those provided in the script (e.g. if
the data ends in .asdm.sdm then move to names which do not have this suffix) • Start the version of CASA containing Pipeline using casapy -‐-‐pipeline, then type
execfile(‘casa_pipescript.py’)
Running the script will create: • a calibrated MS for each ASDM in the same directory • test images of the bandpass and phase calibrator (1 per spectral window in *.image format. To view
a *.image file e.g. use casaviewer image_file_name) • a pipeline*/html directory containing
o the Pipeline WebLog -‐ the html pages of plots from the calibration process. Access using e.g. firefox index.html
o casa_commands.log –a list of the CASA tasks used by Pipeline tasks during calibration
Note that to re-‐run the Pipeline multiple times, it is recommended to start each time from a clean directory containing only the raw data, flux.csv and *flagtemplate.txt files and the casa_pipescript.py script. If the flux.csv and *flagtemplate.txt files are not present in the directory, Pipeline will create new default versions of these files, which will not contain any edits made to them by ALMA staff during the Quality Assurance Process. Also note that it is advisable to have 8-‐10 GB RAM and 50-‐75 GB disk space per ALMA raw data file (ASDM) available to perform the Pipeline calibration. Please contact ALMA via the Helpdesk if assistance is needed with data reprocessing.
6
5.2 Modifying the Pipeline Run
5.2.1 Note on the Pipeline Mode
# Example Pipeline Data Reduction Script (casa_pipescript.py)
__rethrow_casa_exceptions = True
h_init()
try:
hifa_importdata(vis=['uid___A002_X72e960_X53d', 'uid___A002_X72e960_X53f'], session=[‘session_1’, ‘session_1’])
hifa_flagdata(pipelinemode="automatic")
hifa_fluxcalflag(pipelinemode="automatic")
hif_refant(pipelinemode="automatic")
hifa_tsyscal(pipelinemode="automatic")
hifa_tsysflag(pipelinemode="automatic")
hifa_wvrgcalflag(pipelinemode="automatic")
hif_lowgainflag(pipelinemode="automatic")
hif_setjy(pipelinemode="automatic")
hif_bandpass(pipelinemode="automatic")
hif_bpflagchans(pipelinemode="automatic")
hifa_gfluxscale(pipelinemode="automatic")
hifa_timegaincal(pipelinemode="automatic")
hif_applycal(pipelinemode="automatic")
hif_makecleanlist(intent='PHASE,BANDPASS,CHECK')
hif_cleanlist(pipelinemode="automatic")
finally:
h_save()
Figure 1: Example Pipeline Calibration Script
7
In the Pipeline standard calibration script, the Pipeline Mode is defined as “automatic” for each task. In this mode, the task takes the default settings of Pipeline and only a limited number of parameters are exposed for editing by a user. Setting the pipeline mode to “interactive” will usually enable the values of a larger number of parameters to be changed. See the ALMA Science Pipeline Reference Manual for more details.
5.2.2 Removing ASDMs from the processing
ASDMs can be removed from the processing by editing the vis= and session= lists in hifa_importdata in casa_pipescript.py.
5.2.3 Introducing additional flagging
Additional manual flagging can be introduced to any Pipeline reduction by editing the *flagtemplate.txt files. Examples of the syntax to use in editing these files are given at the top of the files. The flags are applied when hifa_flagdata is run.
5.2.4 Editing calibrator fluxes used by the Pipeline
Pipeline currently reads quasar calibrator fluxes from the flux.csv file. If a regularly-‐monitored quasar has been used as the flux calibrator, the flux of this provided from the ALMA Quality Assurance process can be over-‐ridden by editing it in this file. Only values for the flux calibrator can be over-‐ridden, since values for the bandpass and phase calibrators are derived during data calibration
6 The Pipeline WebLog The WebLog is a set of html pages that give a summary of how the calibration of ALMA data proceeded. The WebLog produced during the ALMA Quality Assurance Process will be in the qa directory of an ALMA delivery. To view the WebLog, untar and unzip the file using e.g. tar zxvf *weblog.tar.gz . This will provide a pipeline*/html directory containing the WebLog, which can be viewed using a web browser e.g. firefox index.html. The WebLog aims to provide both a quick overview of datasets and also give methods for exploring the calibration and flagging in detail. Therefore most calibration pages of the WebLog will first give a single “representative” view of that calibration (or flagging) step. Users can then choose whether to further select a detailed view of all the plots associated with that calibration step. These plots can easily be filtered by a combination of outlier, antenna and spectral window criteria. Calibration and flagging steps also undergo an automated scoring to give an “at a glance” indication of any trouble points. Throughout the WebLog, links are denoted by text written in blue and it is usually possible to click on thumbnail plots to enlarge them. Where histograms are displayed, in modern web browsers it is possible to draw boxes on multiple histograms to select the plots associated with those data points.
8
Figure 2: WebLog Home Page. The naviation bar is circled in red.
6.1 Navigation To navigate the main pages of the WebLog, click on items given in the bar at the top of the WebLog home page. Also use the Back button provided at the upper right on some of the WebLog sub-‐pages. Avoid using “back/previous page” on your web browser.
6.2 Home Page The first page in the WebLog gives an overview of the observations (proposal code, data codes, PI, observation start and end time), a pipeline execution summary, and an observation summary. Clicking on the bar at the top of the home page enables navigation to By Topic or By Task.
6.2.1 Observation Summary The Observation Summary table lists all the measurement sets included in the pipeline processing. Each measurement set is calibrated independently by the pipeline. The table provides a quick overview of the ALMA receiver band used, the number of antennas, the start/end date and time, the time spent on source, the array minimum and maximum baseline length, the rms baseline length and the size of that measurement set. To view the observational setup of each measurement set in more detail, click on the name of it to go to its overview page.
6.2.1.1 Measurement set overview pages Each measurement set overview page has a number of tables: Observation Execution Time, Spatial Setup, Antenna Setup, Spectral Setup and Sky Setup. For more information on the tables titled in blue text, click on these links. There are additionally links to Weather and Scans information. Two thumbnail plots, which can be enlarged by clicking on them, show the observation structure either as Field Source Intent vs Time or Field Source ID vs Time. To view the CASA listobs output from the observation click on listobs output.
9
Figure 3: A Measurement Set Overview Page. Click on the table headings in blue to see more information about each.
6.3 By Topic The By Topic page provides an overview of all Warnings and Errors triggered, a Quality Assessment overview in Tasks by Topic and Flagging Summaries for the processing.
6.4 By Task The By Task Summary page gives a list of the calibration and flagging steps performed on the dataset. It is not displayed per measurement set as the Pipeline performs each step on every measurement set sequentially before proceeding to the next step e.g. it will import and register all measurement sets with the Pipeline before proceeding to perform the ALMA deterministic flagging step on each measurement set. The name of each step on the By Task page is a link to more information. On the right hand side of the page, the bars and scores indicate how well the Pipeline processing of that stage went. All green bars should indicate a fairly problem-‐free dataset.
6.4.1 Task sub-‐pages The task sub-‐pages (accessed by clicking on e.g. hifa_importdata) provide the outcome, or the representative outcome, of each Pipeline task executed. Additionally by clicking on the appropriate blue links on each sub-‐page, the CASA Log, Pipeline QA, Input Parameters and Task Execution Times can be accessed. Most sub-‐pages have further links in order to access a more detailed view of the outcome of each task. These links are often labelled by the measurement set name. For a fast assessment of results, go straight to the By Task > hif_applycal page.
10
6.5 Selecting and filtering plot displays Using the By Task > hifa_tsysflag: Flag Tsys calibration pages as an example.
Figure 5: The Tsys plots for all antennas. Above the plots are histograms indicating the distributions of metrics based on the median Tsys.
Figure 4: Flag Tsys Calibration Overview Page. A representative view of the Tsys measurements for each spectral window is shown. To see the measurements for individual antenna/spectral windows, click on the measurement set name, circled here in red.
11
Figure 6: Flitering to the plot of interest by drawing grey boxes on two of the histograms using a mouse button, and adding additional spectral window and antenna filters. To clear the grey box filters on the histograms, click on white space in the histograms.
6.6 Weblog Quality Assessment (QA) Scoring Pipeline tasks have scores associated with them in order to quantify the quality of the dataset and the calibration. The scores are between 0.0 and 1.0 and are colourized such that:
Score Colour Comment 0.90-‐1.00 Green Standard/Good 0.66-‐0.90 Blue Below standard 0.33-‐0.66 Yellow Warning 0.00-‐0.33 Red Error
Pipeline Task Pipeline QA Scoring Metric Score
hifa_importdata Checking that the required calibrators are present
1.0 all present
0.1 subtracted for missing bandpass or flux calibrator
1.0 subtracted for missing phase calibrator or Tsys calibration
0.5 subtracted for existing processing history
12
hifa_flagdata Determining percentage of incremental flagging
Shadowing: 0 < score < 1 = 50% < fraction flagged < 20%
Others: 0%-‐5% -‐> 1.0, 5%-‐50% -‐> 1.0...0.5, >50% -‐> 0.0
hifa_fluxcalflag Determining percentage of incremental flagging
0%-‐5% -‐> 1.0, 5%-‐50% -‐> 1.0...0.5, >50% -‐> 0.0
hif_refant Determining if a reference antenna could be found
1.0 if reference antenna could be determined, 0.0 otherwise
hifa_tsysflag Determining percentage of incremental flagging
0%-‐5% -‐> 1.0, 5%-‐50% -‐> 1.0...0.5, >50% -‐> 0.0
hifa_wvrgcalflag Checking phase RMS improvement
0.0 if RMS(before)/RMS(after) < 1, 0.5 ... 1.0 for ratios between 1 and 2, and 1.0 for ratios > 2
hif_lowgainflag Determining percentage of incremental flagging
0%-‐5% -‐> 1.0, 5%-‐50% -‐> 1.0...0.5, >50% -‐> 0.0
hif_bandpass Judging phase and amplitude solution flatness per antenna, spectral window and polarization (interim measure, future scoring will be based on data with solutions applied)
two algorithms: Wiener entropy and derivative deviation, and signal-‐to-‐noise ratio (scores: Wiener entropy: error function with 1-‐sigma deviation of 0.001 from 1.0; derivative deviation: error function with 1-‐sigma deviation of 0.03 for the outlier fraction; signal-‐to-‐noise ratio: error function with 1-‐sigma deviation of 1.0 for the signal-‐to-‐noise ratio)
hifa_gfluxscale Determining SNR of fitted flux values
fitted flux values with SNR < 5.0 are assigned a score of 0.0, SNR > 20.0 a score of 1.0, and a linearly scaled value in between
hifa_timegaincal Determining X-‐Y / X2-‐X1 phase solution deviations
Standard deviation of X-‐Y phase difference converted to path length: 1.0 if lower than 4.25e-‐6 m, 0.0 if higher than 7.955e-‐2 m, with an exponential decrease in between. Standard deviation of X2-‐X1 phase differences of subsequent integrations converted to path length: 1.0 if lower than 3.08-‐e-‐5 m, 0.0 if higher than 2.24e-‐2 m, with an exponential decrease in between. NB: The high limits are currently dummies. Determining their realistic values is still under development.
hifa_applycal Determining percentage of incremental flagging
0%-‐5% -‐> 1.0, 5%-‐50% -‐> 1.0...0.5, >50% -‐> 0.0
13